Reliability of the Social Phobia and Anxiety Inventory (SPAI) Portuguese version in a heterogeneous sample of Brazilian university students

Prévia do material em texto

Reliability of the Social Phobia and Anxiety Inventory
(SPAI) Portuguese version in a heterogeneous sample
of Brazilian university students
Fidedignidade da versão em português do Inventário
de Ansiedade e Fobia Social (SPAI) em amostra
heterogênea de estudantes universitários brasileiros
AbstractAbstractAbstractAbstractAbstract
Objective:Objective:Objective:Objective:Objective: Tto evaluate internal consistency and temporal stability trough test re-test reliability of the Portuguese version of the Social Phobia
and Anxiety Inventory (SPAI) in a heterogeneous Brazilian sample. MethodsMethodsMethodsMethodsMethods: After informed consent, the translated and adapted version of the
SPAI Portuguese was tested and re-tested at a 14-day interval in a sample of 225 university students, both genders, from 4 different
undergraduate courses. After double data entry, statistical analysis included Cronbach’s alphas, Pearson and intra-class correlation coefficients.
ResultsResultsResultsResultsResults: The studied sample consisted of 213 students from 4 sub-samples: 95 law students, 31 social communication students, 54 civil
engineering students and 33 dental medicine students. The mean age was 23 (±6) years and 110 (51.6%) were female. The Cronbach’ alpha
was 0.96, no difference for both genders. The differential (total) score of the SPAI Portuguese showed Pearson coefficient 0.83 (95% CI 0.78-0.87)
and Intra-class coefficient 0.83 (95% CI 0.78-0.86). There was no statistical difference in Pearson coefficients between genders (p = 0.121)
and the 4 sub-samples (p = 0.258). ConclusionConclusionConclusionConclusionConclusion: The results do not ensure validity, however, the SPAI Portuguese version presented a good
content homogeneity with satisfactory level of internal consistency. The temporal stability evaluated was substantial. The results showed that
the Portuguese SPAI presents perfectly acceptable reliability for both genders and suggest that it can be used in Brazilian populations
.
Keywords: Keywords: Keywords: Keywords: Keywords: Phobic disorders/diagnosis; Reproducibility of results; Self assessment (Psychology); Anxiety disorders/diagnosis; Behavior
therapy/methods
ResumoResumoResumoResumoResumo
ObjetivoObjetivoObjetivoObjetivoObjetivo: Avaliar consistência interna e estabilidade temporal, através de fidedignidade teste-reteste, da versão em português do Inventário
de Ansiedade e Fobia Social (SPAI), em uma amostra brasileira heterogênea. MétodosMétodosMétodosMétodosMétodos: Após consentimento informado, a versão traduzida
e adaptada para o português do SPAI foi testada e re-testada, com 14 dias de intervalo, em uma amostra de 225 estudantes universitários
de ambos os gêneros, de quatro cursos de graduação. Após dupla entrada de dados, a análise estatística incluiu Alfa de Cronbach e os
coeficientes de correlação de Pearson e Intra-classe. ResultadosResultadosResultadosResultadosResultados: A amostra estudada consistiu de 213 estudantes de quatro sub-amostras:
95 estudantes de Direito, 31 de Comunicação Social, 54 de Engenharia Civil e 33 de Odontologia. A média de idade foi de 23 anos (± 6)
e 110 (51,6%) eram do gênero feminino. O Alpha de Cronbach foi de 0,96, sem diferenças entre os gêneros. O escore diferencial (total)
do SPAI português apresentou coeficiente de Pearson de 0,83 (IC95% 0,78-0,87) e coeficiente Intra-classe de 0,83 (IC95% 0,78-0,86).
Não houve diferença estatística nos coeficientes de Pearson entre os gêneros (p = 0,121) ou entre as quatro sub-amostras (p = 0,258).
ConclusãoConclusãoConclusãoConclusãoConclusão: Os resultados não asseguram validade; a versão do SPAI português apresentou boa homogeneidade de conteúdo com nível
satisfatório consistência interna. A estabilidade temporal avaliada foi consistente. Os dados demonstram que o SPAI português apresenta
fidedignidade perfeitamente aceitável para ambos os gêneros e sugerem sua utilização na população brasileira.
Descritores:Descritores:Descritores:Descritores:Descritores: Transtornos fóbicos/diagnóstico; Reprodutibilidade de resultados; Auto-avaliação (Psicologia); Transtornos da ansie-
dade/diagnóstico; Terapia comportamental/métodos
1 Post-graduate Program in Medical Sciences: Psychiatry - Universidade Federal do Rio Grande do Sul (Federal University of Rio Grande
do Sul, UFRGS), Porto Alegre, RS, Brazil
2 Department of Psychiatry and Legal Medicine of the Pontifícia Universidade Católica do Rio Grande do Sul (Pontifical Catholic
University of Rio Grande do Sul, PUCRS). Porto Alegre, RS, Brazil
3 Research and Post-graduate Group of the Hospital de Clínicas de Porto Alegre, RS, Brazil
4 Department of Psychology, University of Maryland. Baltimore, MD, USA.
Rev Bras Psiquiatr. 2005;27(2):124-30
Financing: This work was partly supported by The Research
Foundation of the State of Rio Grande do Sul
(FAPERGS; nº 97/50734.9).
Conflicts of interests: None
Submitted: 21 September 2004
Accepted: 25 November 2004
Correspondence
Patrícia Picon
Rua Padre Chagas 415/803
90570-080 Porto Alegre, RS, Brasil
Phone: (55 51) 3346-9022
E-mail: ppicon@terra.com.br
ORIGINAL ARTICLEORIGINAL ARTICLEORIGINAL ARTICLEORIGINAL ARTICLEORIGINAL ARTICLE
124
Original version accepted in EnglishOriginal version accepted in EnglishOriginal version accepted in EnglishOriginal version accepted in EnglishOriginal version accepted in English
Patrícia Picon,1,2 Gabriel J C Gauer,2 Vânia N Hirakata,3
Lourdes M Haggsträm,2 Deborah C Beidel,4
Samuel M Turner,4 Gisele G Manfro1
This paper was elaborated from data set of the first author Ph. D. degree at Post-graduate Program in Medical Sciences: Psychiatry -
Universidade Federal do Rio Grande do Sul (Federal University of Rio Grande do Sul, UFRGS).
Part of the results were presented as posters at: 28th Annual Meeting of the Brazilian Psychiatric Association, October 25-28, 2000, Rio
de Janeiro, RJ, Brazil and 157th Annual Meeting of the American Psychiatric Association, May 1-6, 2004, New York, NY, USA.
Art06_rev04.p65 13/5/2005, 11:53124
125 Picon P et al
Rev Bras Psiquiatr. 2005;27(2):124-30
agoraphobia.18,20 The construct’s concurrent, convergent,
external and discriminant validities of the SPAI Inventory have
been demonstrated in clinical and non-clinical samples.21-23
The SPAI Inventory appropriately discriminates patients with
social phobia from other anxiety disorders and from normal
controls.2 Its sensitivity to measure therapeutic results has been
proved.24-26
SPAI has translations into at least 10 languages.2 There is
already a study to validate the Spanish version in a non-clinical
sample where the results of internal consistency, test- retest
reliability and factorial analysis were similar to those in the
original English version.27-28
The lack of instruments to screen and evaluate the intensity
of anxiety or social phobia symptoms adapted to the Portuguese
language and Brazilian culture, with appropriate measurement
of psychometric properties, has been felt as a serious gap.
The validation of the SPAI Portuguese version would make it
possible to design studies to identify probable cases, evaluate
the severity of symptoms and follow up therapeutic results in
social phobia in Brazilian samples of adults and adolescents
aged 14 years or over. These studies would be comparable to
those of other countries, populations and cultures, and thus
of interest to the Brazilian clinical and scientific community.
Furthermore, the need of further research to broaden the SPAI
validation process in non-American samples, has been already
indicated in the literature, as a form of confirming and
expanding its generalization.29
The aims of this study were to evaluate internal consistency
and temporal stability, using the test-retest reliability measure
of the Portuguese version of the Social Phobia and Anxiety
Inventory (SPAI)21,30 in a heterogeneous population sample of
Brazilian university students.
MethodsMethodsMethodsMethodsMethods
1. Sample1. Sample1. Sample1. Sample1. Sample
Using a cross-sectionaldesign, 213 university students of
both genders were evaluated and re-evaluated to the study of
internal consistency and test-retest reliability. The study included
fresh and senior students of 9 classes allotted from four different
undergraduate courses. Three hundred and twenty-six individuals
were assessed in the first evaluation aiming to assure minimal
sample size of 200 participants in the second evaluation,31 taking
into account that students might be absent or refuse to
participate. During the second evaluation 94 were absent and
there were 7 refusals. Thus 225 students participated in the
re-test. Among the 225 participants who accepted to participate,
12 (5.3%) had problems to complete the SPAI, either at the
first or the second evaluation, which made it impossible to
calculate the (total) differential score.18 The final sample studied
then consisted of 213 students from 4 sub-samples: 95 law
students, 31 social communication students, 54 civi l
engineering students and 33 dental medicine students.
2. Procedures2. Procedures2. Procedures2. Procedures2. Procedures
After the formal consent of the authors and the publisher,
Mult i -Heal th Systems Incorporat ion, which holds the
copyrights, the original Social Phobia Anxiety Inventory (SPAI)
English version18 was translated into Portuguese by Picon and
Gauer in 1999.30
The face and content validity was performed by a translation
and back-translation process, evaluation study with bilingual
volunteers and evaluation of the Portuguese version by a panel
of experts.32
IntroductionIntroductionIntroductionIntroductionIntroduction
Social anxiety disorder or social phobia is a relatively recent
diagnostic category.1 It presents lifetime prevalence rates
ranging from 7 to 16% in epidemiological studies of the
American and European population.2 It affects very young
individuals with a mean age of onset between 15 and 18
years, a chronic course and high comorbidity.3-5 However,
although this disorder responds to currently available
pharmacological and psychotherapeutic treatments,6-10 only
25% of social phobics receive appropriate treatment.11 Early
psychiatric diagnosis depends on a high degree of suspicion.
Screening for social anxiety disorder should be part of the
routine in any psychiatric evaluation.12 The detection of
probable cases of social anxiety disorder using screening
instruments and the evaluation of the intensity of social anxiety
symptoms in a clinical and research environment requires
valid and reliable instruments.13-15
Although in psychiatric clinical work it is traditional to
emphasize measures or assessments performed by clinical
evaluators, in social anxiety disorder the self-reported measures
are more widely used and more extensively validated.7-9 Several
of these instruments are currently available, being specifically
developed to evaluate social phobia, with acceptable
psychometric properties, and have been tested in clinical
samples. They have been used as screening instruments to
identify probable cases and evaluate the severity of symptoms
in therapeutic effectiveness studies. Outstanding among these
scales is the Social Phobia and Anxiety Inventory (SPAI), the
most extensive self-assessment instrument that presents an
excellent psychometric profile and has demonstrated its capacity
to evaluate treatment outcome. It was validated both in
population and clinical samples, in adults and adolescents
aged 14 years or over. Its format allows its use for therapeutic
planning.16 The SPAI has a subscale on agoraphobia that
enables the exclusion of cases with panic disorder and
agoraphobia as a differential diagnosis.3,17
Social Phobia and Anxiety InventorySocial Phobia and Anxiety InventorySocial Phobia and Anxiety InventorySocial Phobia and Anxiety InventorySocial Phobia and Anxiety Inventory
The Social Phobia and Anxiety Inventory (SPAI) was developed
by Turner and co-workers in 1989 to screen for probable ca-
ses of Social Phobia, evaluate the severity of symptoms of
social anxiety and therapeutic results. The SPAI inventory looks
at physiological, behavioral and cognitive aspects of the
dimensional construct of anxiety and social phobia.18
The SPAI inventory consists of 45 items that quantitatively
evaluate the severity of the symptoms of anxiety and social
avoidance using a 7-point Likert scale. It has two subscales:
social phobia (items 1 to 32) and agoraphobia (items 33 to
45). The differential or total score of the SPAI is calculated by
subtracting the agoraphobia score from the social phobia score.
This differential (total) score of the SPAI is indicated as the
most accurate to discriminate between cases and non-cases
of probable social phobia in clinical and non-clinical studies,
minimizing the number of false positives or negatives.18
The SPAI Inventory may then be used as a screening
instrument, with good discriminatory performance and a cutoff
point equal to or greater than 80 to identify probable cases of
social phobia.19
The original English version was tested in population samples
of American university students, and adolescents, and in
clinical samples. SPAI presented excellent internal consistency
and test-retest reliability in these studies. The factorial analyses
showed the presence of two factors: social phobia and
Art06_rev04.p65 13/5/2005, 11:54125
Rev Bras Psiquiatr. 2005;27(2):124-30
Reliability of the SPAI Portuguese version in a Brazilian sample 126
correlation coefficient and intra-class correlation coefficient
with their respective 95% confidence intervals.31,35
The comparisons of Pearson correlation coefficient between
genders, and between the four different sub-samples of
undergraduate courses were performed using the Chi-square
heterogeneity test.36
ResultsResultsResultsResultsResults
In the total sample of 213 participants, the mean age was
23 ± 6 years. The minimum age was 17, the maximum 53,
with a median of 22 years. One hundred and three students
were male (48.4%) and 110 female (51.6%). There was no
statistical difference (p = 0.832) in the mean age between
males (23 ± 5.4 y) and females (23 ± 6.5 y).
The Shapiro-Wilk normality test revealed asymmetry in the
social phobia and agoraphobia scores, which underwent square
root transformation. The results of analyses with and without
transformation were similar and the later are presented.
The means of the social phobia, agoraphobia and differential
(total) SPAI Portuguese scores obtained at the first evaluation
were not different between genders (Table 1) or between the
four different sub-samples of undergraduate courses studied (Table
2). The total sample (n = 213) was therefore homogeneous in
relation to the means of the scores that are being studied.
The Internal Consistency estimated using Cronbach’s alpha
in the total sample and in both genders were identical for the
social phobia subscale (items 1 to 32; alpha = 0.96) and for
the total scale SPAI Portuguese (items 1 to 45; alpha = 0.96).
The agoraphobia subscale (items 33 to 45) presented alpha
values of 0.87 in the total sample, 0.88 in the female group
and 0.85 in the male group. The Cronbach’s alpha coefficients
of the two subscales and of the SPAI Portuguese were similar
in the four sub-samples, except for the agoraphobia subscale,
in which it presented the greatest variation (Table 3).
The test-retest reliability of social phobia, agoraphobia and
differential (total) SPAI Portuguese scores, estimated using
The SPAI Portuguese version also consisted of two subscales:
social phobia and agoraphobia, with a total of 45 items, scored
using a 7-point Likert scale, similar to the English version.30
After the approval by the São Lucas Hospital’s Ethics in Research
Committee at Pontifical Catholic University of Rio Grande do Sul,
data collection was approved by the directors of the college courses
allotted for the implementation of this study. After the end of
routine academic activities, and with the verbal consent of the
teachers responsible for the classes studied,the students in the
classroom, who agreed to participate in the project, having signed
the informed consent, completed the SPAI Portuguese.
The SPAI Portuguese was tested and re-tested at a 14-day
interval,31 to all participants in the sample studied. The students
who participated in the temporal stability study received a code
number to perform the re-test.
3. Statistical analysis3. Statistical analysis3. Statistical analysis3. Statistical analysis3. Statistical analysis
The data base was performed by independent double entry and
later checked by the Validate module of EPI INFO version 6.04d
(World Health Organization, Geneva). The statistical analyses
were performed using SPSS version 10.0 (SPSS Inc., Chicago),
STATA version 7.0 (STATA Corp., College Station), and PEPI version
3.0 (PEPI, Station Building, UK) software. The scores of social
phobia, agoraphobia subscales and of the SPAI Portuguese
differential (total) score were calculated using SPSS version 10.0
and tested as to their normality by Shaphiro-Wilk test.33
The comparisons of age means and the scores of social
phobia, agoraphobia and differential (total) SPAI Portuguese
between the four courses were performed using one-way
analysis of variance (ANOVA).
Internal consistency for the items of the subscales of social
phobia and agoraphobia and for all items of the SPAI Portuguese
was evaluated using Cronbach’s alpha coefficient.34 The test-
retest correlation and concordance estimates for the social
phobia and agoraphobia scores and for the differential (total)
score of the SPAI Portuguese were evaluated using Pearson
Art06_rev04.p65 13/5/2005, 11:54126
127 Picon P et al
Rev Bras Psiquiatr. 2005;27(2):124-30
Pearson correlation coefficients in the four sub-samples (Table
3) in the total sample and in both genders (Table 4) varied
from: 0.82 to 0.93 for the social phobia score; 0.72 to 0.85
for the agoraphobia score; and 0.79 to 0.90 for the differential
(total) SPAI Por tuguese score. The Pearson correlation
coefficients did not present a significant difference in the
comparison between the four sub-samples (p = 0.833) or
between genders (p = 0.740) for the social phobia score.
The Pearson correlation coefficients also did not present a
significant difference in the comparison between the four sub-
samples (p = 0.258), or between genders (p = 0.121) for
the differential (total) SPAI Portuguese score. However, there
was a significant difference in the comparison of the Pearson
correlation coefficients between genders for the agoraphobia
score (p = 0.029) that did not repeat in the comparison
between the sub-samples (p = 0.713).
The test-retest reliability of the three scores studied, estimated
using the intra-class correlation coefficient in the four sub-
samples (Table 3), in the total sample and in both genders
(Table 4) varied from: 0.81 to 0.92 for the social phobia score;
0.68 to 0.85 for the agoraphobia score; and 0.79 to 0.90 for
the (total) differential SPAI Portuguese score.
DiscussionDiscussionDiscussionDiscussionDiscussion
The present study suggests that, in the sample studied, the
Portuguese version of the Social Phobia and Anxiety Inventory
(SPAI) presented results equivalent to the original English
version as to its psychometric properties of internal consistency
and test-retest reliability.
The analyses performed took into account the distribution of
the variables studied, as previously described. Only after careful
analysis of the data with and without normalization of the
social phobia and agoraphobia scores a decision was made to
use the analyses without transformation. The results are
compared to those obtained in American and Spanish
population samples of young adults and adolescents.
Means of the social phobia and agoraphobia scores andMeans of the social phobia and agoraphobia scores andMeans of the social phobia and agoraphobia scores andMeans of the social phobia and agoraphobia scores andMeans of the social phobia and agoraphobia scores and
differential (total) SPdifferential (total) SPdifferential (total) SPdifferential (total) SPdifferential (total) SPAI PAI PAI PAI PAI Pororororortuguese scoretuguese scoretuguese scoretuguese scoretuguese score
The current total sample (n = 213) consisted mostly of young
adults. The means of the social phobia scores and the (total)
differential SPAI Portuguese score, in the four sub-samples,
in both genders and in the total sample studied were greater
then those found in population studies of Spanish adolescents
and American university students and adolescents without
Art06_rev04.p65 13/5/2005, 11:55127
Rev Bras Psiquiatr. 2005;27(2):124-30
Reliability of the SPAI Portuguese version in a Brazilian sample 128
social anxiety.18,20,27 The means of the agoraphobia scores found
in the current study were very similar to those found in the
American studies.
In a study with American adolescents, in the group without
socia l anxiety, the mean score of socia l phobia and
agoraphobia, and of the differential (total) SPAI score were
59.7 ± 31.7, 17.3 ± 11.9 and 43.0 ± 26.4, respectively.
These means were statistically lower than the means of the
social phobic group.20
The mean of the differential (total) SPAI score in a sample of
American psychology students with social phobia (n = 58)
was 72.2 ± 20.4 and among students without social anxiety
(n = 124) it was 32.7 ± 21.3.21 In the group with social
anxiety the mean of the social phobia score and of the
differential (total) SPAI score were higher among male
individuals. The mean agoraphobia score was not different
between genders.18,22
In the present study, the means of the three scores did not
present differences between genders, as opposed to what was
observed in the American as well as in the Spanish studies.
In the American study with a smaller sample size, in the group
without social anxiety, only the agoraphobia score was the
same between genders, and the others were higher among
males.22 In the Spanish study, in the total sample of 3160
adolescents from 14 to 17 years, girls had higher means in
all three scores.27
The differences in the means of social phobia and differential
(total) SPAI Portuguese scores, and the differences of means
between genders as compared to the American and Spanish
samples may be attributed to: the Brazilian sample does not
discriminate students with and without social anxiety, different
sample sizes, and cultural differences in the way of expressing
social anxiety and agoraphobia symptoms among genders.
Of note, similarities of the three mean scores in the four
sub-samples and between genders in the total sample (n =
213) give us indirect evidence of homogeneity of the final
sample, as regards to a set of symptom expression of social
anxiety and agoraphobia in our culture, even between groups
with supposedly distinct profiles of professional choices.
Internal consistency of social phobia and agoraphobiaInternal consistency of social phobia and agoraphobiaInternal consistency of social phobia and agoraphobiaInternal consistency of social phobia and agoraphobiaInternal consistency of social phobia and agoraphobia
subscales and of SPsubscales and of SPsubscales and of SPsubscales and of SPsubscales and of SPAI PAI PAI PAI PAI Pororororortuguesetuguesetuguesetuguesetuguese
In the present study the results of Cronbach’s alpha
coefficients for the subscales of social phobia and agoraphobia
and the total SPAI Portuguese version showed internal levels
of consistency that are satisfactory to excellent and were simi-
lar in the four sub-samples.15,31 However, for the agoraphobia
subscale Cronbrach’s alpha coefficient presented a small
variation. In the four sub-samples the values obtained for the
social phobia and agoraphobia subscales and the total SPAI
Portuguese version were very similar or slightly higher then
those obtained in the American and Spanish studies.
 In the American sample of psychology students with a mean
age of 19 years Cronbach’s alpha coefficients were:0.96 for
the social phobia subscale and 0.85 for the agoraphobia
subscale.18,21-22 In the American sample of 223 adolescents
from 12 to 18 years, the Cronbach’s alpha coefficients were:
0.97 for the social phobia subscale and 0.91 for the
agoraphobia subscale and total SPAI Inventory.20 In the Spanish
study, in a sample of 3160 adolescents, aged 14 to 17,
Cronbach’s alpha coefficients were: 0.95 for the social phobia
subscale; 0.84 for the agoraphobia subscale and 0.95 for the
total SPAI Inventory.27
In the current total sample, the internal consistency results
were the same between genders for the social phobia subscale
(0.96), identical to the results in a sample of American
university students. In the agoraphobia subscale, Cronbach’s
alpha coefficient was 0.88 among the women and 0.86
among the men, and in the American sample the alpha values
were the same in both genders (0.85).18,21-22 The Cronbach’s
alpha coefficient for the total SPAI Portuguese was also identical
for both genders (0.96), but there are no comparable data in
the American or Spanish samples according to gender.18,28
The results showed that social phobia and agoraphobia
subscales were equally reliable for men and women, as the
result obtained in an American study. The total SPAI Portuguese
Inventory also was equally reliable for men and women in this
Brazilian student sample.
The findings reveal that the SPAI Portuguese version presents
a good content homogeneity expressed by the correlation
between its items, measured by the Cronbach’s alpha
coefficient. The values of Cronbach’s alpha coefficient of the
two subscales and the total SPAI Portuguese scale showed a
high correlation between the items that evaluate the two
constructs of the SPAI Inventor y: socia l phobia and
agoraphobia.15,31,34 These results may be an indirect evidence
of how well the translation and back-translation process were
performed. The process appears to have preserved the
dimensions contained in the original SPAI inventory, with a
possible equivalent content, to be appropriately investigated in
an ongoing content validation study using factorial analysis.28
TTTTTest-retest reliability for social phobia and agoraphobiaest-retest reliability for social phobia and agoraphobiaest-retest reliability for social phobia and agoraphobiaest-retest reliability for social phobia and agoraphobiaest-retest reliability for social phobia and agoraphobia
scores and the differential (total) score of the SPscores and the differential (total) score of the SPscores and the differential (total) score of the SPscores and the differential (total) score of the SPscores and the differential (total) score of the SPAI PAI PAI PAI PAI Pororororortuguesetuguesetuguesetuguesetuguese
Pearson correlation coefficients for the social phobia and
agoraphobia scores and differential (total) score of the SPAI
Portuguese in the four sub-samples and in the total sample
were similar to those of the American sample of 182 university
students. In the American study the Pearson correlation
coefficients were: 0.85 for the social phobia score; 0.74 for
the agoraphobia score; and 0.86 for the differential (total)
score of the SPAI, which presented no difference between
genders, being 0.87 between males and 0.85 between
females.18,21
In the total sample Pearson correlation coefficients were:
0.88 for the social phobia score; 0.80 for the agoraphobia
score and 0.83 for the differential (total) score of SPAI
Portuguese. For the differential (total) score, the difference
between genders in the Brazilian sample (see Table 4) did not
present a significant difference (p = 0.163), as in the
American study.18-21
The data reveal identical temporal stability for the social
phobia subscale and the SPAI Portuguese inventory for both
genders and for the four groups studied.
 The Pearson correlation coefficient in the total sample for
agoraphobia score was 0.85 for females and 0.74 for males.
Females show a higher temporal stability (p = 0.029), than
males for agoraphobia items. It will be difficult to explain this
difference as due to sample size, because no differences were
detected between the four sub-samples that have a smaller
number of individuals.
Concordance analyses through intra-class correlation
coefficient were also performed, since this is the most appropriate
form of evaluation of the temporal stability of an instrument
whose score is treated as a continuous variable.28,37 As to the
test-retest reliability estimates, the intra-class correlation controls
Art06_rev04.p65 13/5/2005, 11:55128
129 Picon P et al
Rev Bras Psiquiatr. 2005;27(2):124-30
the effect of chance over variance between individuals and
between test and retest more appropriately, when compared to
the Pearson correlation coefficient method.15,31
Intra-class correlation analyses showed results that were
mostly identical or very similar to those of the analyses of
Pearson correlation coefficients, in the four sub-samples, in
both genders and in the total sample. These findings may be
attributed to the adequate size of the sample and to the score
variations between the test and retest being due mainly to
biological variability and to the difference inherent to the
passage of time between the two assessments, which may
indicate the absence of measurement bias. The isolated use
of the Pearson correlation coefficient may overestimate the
results of reliability because it does not take into account a
possible presence of systematic measurement errors.28,35,38
The 95% confidence intervals of the Pearson correlation
coefficients and the intra-class correlations for the three scores
studied in the four sub-samples were rather narrow, once again
indicating an appropriate sample size. The lower limits were
mostly above 0.70 (Tables 3 and 4). In the total sample the
95% confidence intervals of results for r and ICC are always
above 0.75 (Table 3).
The findings are compatible with test-retest correlation and
concordance indexes from good to excellent and perfectly
acceptable for the three scores: social phobia, agoraphobia
and differential (total) score of the SPAI Portuguese.
Although good results for instrument reliability do not assure
its validity, instrument validity is unfeasible when reliability
results are poor.28,38 The results presented here indicate
appropriate reliability to use the SPAI Portuguese version in
Brazilian population samples. However, further studies are
still needed to show their construct validity in Brazilian
population and clinical samples.
Thus the results do not assure the validity of the SPAI
Portuguese version, currently under investigation in factorial
validity studies in a population sample and as to its convergent
and discriminant validities in a clinical sample. Therefore,
this study is part of the broader validation process of the
Portuguese language version of the Social Phobic and Anxiety
Inventory in Brazilian samples.
ConclusionsConclusionsConclusionsConclusionsConclusions
The results in the four sub-samples and in the total sample
studied show fully satisfactory levels of internal consistency of
the social phobia and agoraphobia subscales and Portuguese
language SPAI Inventory, with a substantial temporal stability of
the social phobia, agoraphobia scores and differential (total)
score of the SPAI Portuguese. The data show that the Portuguese-
language version of SPAI shows perfectly acceptable reliability.
These results are comparable to those found in population
samples of American university students and adolescents where
the English language original version was developed.
The heterogeneous sample studied allows us to suggest its
applicability to non-clinical Brazilian populations of both
genders, as well as to heterogeneous samples of over-14-year-
old adolescents and young adults.
The advantages of the self-reported scales, both for clinical
care and for research, are that they are easily applied and
have a low cost. It is unnecessary to train raters, and the
data collectedare easy to compare over time and between
different patients. Its use in different populations, different
therapeutic settings and different cultures will depend on
the capacity to generalize the research data about the
psychometric properties of these instruments. Thus, validation
studies in samples of different cultures increase the external
validity of these instruments.
Due to the family, academic, occupational, social, affective
and sexual environments limitations social phobia imposes on
those who have it, the disorder has been recognized as a
public health problem. The application of the SPAI Inventory
by itself should not be used to confirm the definitive diagnosis
of Social Phobia, and a psychiatric clinical evaluation is
required.18 However, the early identification of probable cases
of social phobia by means of duly validated screening
instruments such as the SPAI Portuguese language version,
reinforces the importance of the data presented here.
AcknowledgmentsAcknowledgmentsAcknowledgmentsAcknowledgmentsAcknowledgments
To Ana Carolina Seganfredo MD, Caroline Dei Ricardi MD, Ana
Carolina Castro MD, Cassiane Bonato MD for their participation on
field work. To Norberto Martins MD, for his cooperation on data
bank work and computerized scores calculations.
ReferencesReferencesReferencesReferencesReferences
1 .1 .1 .1 .1 . American Psychiatric Association. Diagnostic and statistical manual
of mental disorders, 3rd ed. DSM III. Washington (DC): APA; 1980.
2 .2 .2 .2 .2 . Furmark T. Social phobia: overview of community surveys. Acta
Psychiatr Scand. 2002;105(2):84-93.
3 .3 .3 .3 .3 . Beidel DC, Turner SM. Shy children, phobic adults: nature and
treatment of social phobia. Washington (DC): American Psychological
Association; 1998.
4 .4 .4 .4 .4 . Liebowitz MR. Update on the diagnosis and treatment of social
anxiety disorder. J Clin Psychiatry. 1999;60(Suppl 18):22-6.
5 .5 .5 .5 .5 . Lang AJ, Stein MB. Social phobia: prevalence and diagnostic
threshold. J Clin Psychiatry. 2001;62(Suppl 1):5-10.
6 .6 .6 .6 .6 . Pollack MH. Comorbidity, neurobiology, and pharmacotherapy of so-
cial anxiety disorder. J Clin Psychiatry. 2001;62(Suppl 1)2:24-9.
7 .7 .7 .7 .7 . Feske U, Chambless DL. Cognitive behavioral versus exposure only
treatment for social phobia: a meta-analysis. Behav Ther.
1995;26(4):695-720.
8 .8 .8 .8 .8 . Fedoroff IC, Taylor S. Psychological and pharmacological treatments
of social phobia: a meta-analysis. J Clin Psychopharmacol.
2001;21(3):311-24.
9 .9 .9 .9 .9 . Heimberg RG. Cognitive-behavioral therapy for social anxiety disorder:
current status and future directions. Biol Psychiatry. 2002;51(1):101-8.
10.10.10.10.10. Picon P. Terapia cognitivo comportamental do transtorno de ansieda-
de social. In: Caminha RM, Wainer R, Oliveira M, Piccoloto NM,
organizadores. Psicoterapias cognitivo-comportamentais: teoria e
prática. São Paulo: Casa do Psicólogo; 2003. p. 129-44.
11.11.11.11.11. Dingemans AE, van Vliet IM, Couvee J, Westemberg HG.
Characteristics of patients with social phobia and their treatment in
specialized clinics for anxiety disorders in the Netherlands. J Affect
Disord. 2001;65(2):123-9.
12.12.12.12.12. Picon P, Gauer GJC. Fobia social. In: Cataldo Neto A, Gauer GJC,
Furtado NR, organizadores. Psiquiatria para estudantes de medici-
na. Porto Alegre: EDIPUCRS; 2003. p. 422-9.
13.13.13.13.13. Carmines EG, Zeller RA. Reliability and validity assessment. In:
Sullivan JL, editor. Quantitative applications in the social sciences.
Beverly Hills (CA): Sage University Paper; 1979. [Serie 111].
14.14.14.14.14. Peck DF, Dean C. Measurement in psychiatry. In: Kendell RE, Zealley
AK, editors. Companion to psychiatry studies. 3rd ed. Edinburg:
Churchill Livingstone; 1983. p. 223-4.
15.15.15.15.15. Tsuang MT, Tohen M, Zahner GEP, editors. Textbook in psychiatry
epidemiology. New York: Wiley-Liss & Sons; 1995.
16.16.16.16.16. Clark DB, Feske U, Masia CL, Spaulding SA, Brown C, Mammen O,
et al. Systematic assessment of social phobia in clinical practice.
Depress Anxiety. 1997;6(2):47-61.
17.17.17.17.17. Ries BJ, McNeil DW, Boone ML, Turk CL, Carter LE, Heimberg RG.
Assessment of contemporary social phobia verbal report instruments.
Behav Res Ther. 1998;36(10):983-94.
Art06_rev04.p65 13/5/2005, 11:55129
Rev Bras Psiquiatr. 2005;27(2):124-30
Reliability of the SPAI Portuguese version in a Brazilian sample 130
18.18.18.18.18. Turner SM, Beidel DC, Dancu CV. SPAI - Social Phobia & Anxiety
Inventory : manual. Toronto: Multi-Health Systems; 1996.
19.19.19.19.19. Beidel DC, Turner SM. Scoring the social phobia and Anxiety
inventory: comments (technical note). J Psychopathol Behav Assess.
1992;14:377-9.
20.20.20.20.20. Clark DB, Turner SM, Beidel DC, Donovan JE, Kirisci L, Jacob RG.
Reliability and validity of the Social Phobia and Anxiety Inventory
for adolescents. Psychol Assess. 1994;6(2):135-40.
21.21.21.21.21. Turner SM, Beidel DC, Dancu CV, Stanley MA. An empirically derived
inventory to measure social fears and anxiety: the Social Phobia and
Anxiety Inventory. Psychol Assess. 1989;1(1):35-40.
22.22.22.22.22. Beidel DC, Turner SM, Stanley MA, Dancu CV. The Social Phobia
and Anxiety Inventory: concurrent and external validity. Behav Ther.
1989;20:417-27.
23.23.23.23.23. Turner SM, Stanley MA, Beidel DC, Bond L. The Social Phobia and
Anxiety Inventory: construct validity. J Psychopathol Behav
Assess.1989;11:221-34.
24.24.24.24.24. Turner SM, Beidel DC, Long PJ, Turner MW, Townsley RM. A
composite measure to determine the functional status of treated
social phobics: the social phobia endstate functioning index. Behav
Ther.1993;24:265-75.
25.25.25.25.25. Beidel DC, Turner SM, Cooley MR. Assessing reliable and clinically
significant change in social phobia: validity of the social phobia
and anxiety inventory. Behav Res Ther. 1993;31(3):331-7.
26.26.26.26.26. Turner SM, Beidel DC, Wolff PL. A composite measure to determine
improvement following treatment for social phobia: the index of
social phobia improvement. Behav Res Ther. 1994;32(4):471-6.
27.27.27.27.27. Olivares J, García-Lopez LJ, Hidalgo MD, Turner SM, Beidel DC.
The Social Phobia and Anxiety Inventory: reliability and validity in
an adolescent Spanish population. J Psychopathol Behav Assess.
1999;21(1):67-78.
28.28.28.28.28. Rush AJ, Pincus HA, First MB, Blacker D, Endicott J, Keith SJ, et al.
Task force for the Handbook of Psychiatric Measures. Washington
(DC): American Psychiatric Association; 2000.
29.29.29.29.29. McNeil DW, Ries BJ, Turk CL. Behavioral assessment: self-report,
physiology, and overt behavior. In: Heimberg RG, Liebowitz MR,
Hope DA, Schneier FR. Social Phobia, diagnosis, assessment and
treatment. New York: Guilford Press; 1995. p. 202-31.
30.30.30.30.30. Turner SM, Beidel DC, Dancu CV. SPAI- Social Phobia & Anxiety
Inventory. Translated into Portuguese by Picon P, Gauer G. Toronto:
Multi-Health Systems; 1996, 1999. p. 1-6.
31.31.31.31.31. Streiner DL, Norman GR. Health measurement scales - a practical
guide to their development and use. 2nd ed. London: Oxford
University; 1995.
32.32.32.32.32. Picon P, Gauer G, Haggsträm L, Seganfredo A, Dei Ricardi C, Manfro
G. Estudo de confiabilidade da versão em português do Inventário
de Ansiedade e Fobia Social (SPAI) em uma amostra de voluntários
brasileiros bilíngües. Rev Bras Psiquiatr. 2002;24(Supl 2):36.
33.33.33.33.33. Altman DG. Practical statistics for medical research. London:
Chapman and Hall; 1991.
34.34.34.34.34. Cortina JM. What is coefficient Alpha? An examination of theory and
applications. J Appl Psychol. 1993;78(1):98-104.
35.35.35.35.35. Bartko JJ. Measures of agreement: a single procedure. Stat Med.
1994;13(5-7):737-45.
36.36.36.36.36. Zar JH. Biostatistical analysis. 4th ed. New Jersey: Prentice Hall;
1999. p. 377-412.
37.37.37.37.37. Gorenstein C, Andrade LHS, Zuardi AW. Escalas de avaliação clíni-
ca em psiquiatria e psicofarmacologia. São Paulo: Lemos; 2000.
38.38.38.38.38. Streiner DL. A checklist forevaluating the usefulness of rating scales.
Can J Psychiatry. 1993;38(2):140-8.
Art06_rev04.p65 13/5/2005, 11:55130